Substance use disorders are considered complex traits caused by genetic and environmental factors. A new level of understanding of the etiology of these disorders may be achieved by sharpening the research methods used. This includes developing improved phenotypes that are maximally informative and specific. This study will combine an unprecedented set of resources and investigators to develop such phenotypes. Focusing on marijuana, cocaine and nicotine, sophisticated statistical techniques in the area of measurement will be used with existing, ready data, including latent trait modeling with the new release of Mplus and other modeling techniques. These methods will be applied to the largest and most detailed national survey of substance use disorders yet conducted, the 2002 National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), a nationally representative sample of 43,093 respondents. A number of specific binary and continuous traits will be developed and examined for structural invariance over gender, race/ethnicity and age. Importantly, the relationship of these traits to family history variables will also be examined. Areas of attention will include quantitative representation of dependence and withdrawal, dynamic aspects of history and course, commonality vs. specificity of substance disorders, and whether comorbid psychiatric and substance disorders (or some of their components) constitute informative phenotypes. Then, in an innovative link between epidemiology and genetics, the analyses will move to data from the Collaborative Study on the Genetics of Alcoholism (COGA), a large family study of 956 probands and their relatives, of whom 262 probands and their relatives participated in a genetic linkage component. This is also a rich resource of information on drug-abusing individuals since a high proportion of the probands and their relatives used, abused and were dependent on drugs. The latent trait analyses will be redone using COGA interview data. Phenotypes resulting from these analyses will then be investigated in qualitative and quantitative linkage analyses of COGA data. The larger aim is to identify substance phenotypes that can be used in many other studies of genes and their interaction with environmental factors, in order to achieve a better understanding of the etiology of substance use disorders.